Leakage fault detection in Electro-Hydraulic Servo Systems using a nonlinear representation learning approach

Area of Science:

• Mechanical Engineering
• Control Systems
• Machine Learning

Background:

• Electro-Hydraulic Servo Systems (EHSS) are critical actuators in heavy industry.
• EHSS are susceptible to undetected faults like leakage and seal damage.
• These faults degrade performance and can cause system complications.

Purpose of the Study:

• To develop a complexity-reduced method for fault detection in EHSS.
• To utilize representation learning for enhanced fault identification.
• To improve the reliability and performance of industrial EHSS.

Main Methods:

• Employed representation learning to transform raw hydraulic supply pressure signals.
• Developed a custom optimization algorithm with a cost function to find a nonlinear data mapping.
• Validated the proposed transformation in an experimental setup.

Main Results:

• Successfully transformed raw sensor data into a new representation space.
• Demonstrated that the new representation facilitates easier fault classification.
• Experimental validation confirmed the effectiveness of the proposed method.

Conclusions:

• Representation learning offers a promising approach for detecting complex EHSS faults.
• The developed method simplifies fault detection, enhancing system diagnostics.
• This research contributes to improved industrial actuator maintenance and performance.